My Focus Merge output problem

[Salus]

Hi,

I am working on an iMac running Big Sur version 11.3.1 with the Affinity Photo Desktop version 1.9.3 and am having a problem with Focus Merge.

I ran a few different Focus Merges on RAW image files earlier today with no problems. Then I ran a focus merge of 35 raw images of 43MB each and this time I am getting a blur on the left side and bottom of the output composite. When I look at the source files (where the left side of the original image is in focus) the source images show a weird corruption of the left and bottom edges just where the composite is blurred. There is nothing wrong in the original RAW image files that I used to generate the Focus Merge so the problem is created during the Focus Merge process.

I have tried this particular Focus Merge (quitting AF in between trials and selecting the same RAW images) now 3 different times with the same result.

I will include 3 screen grabs to illustrate the issue: one of one of the original RAW images I used, opened in Affinity Photo to show that there is nothing wrong with the file - one of the output of the Focus Merge showing the composite with blurred edges - and the last revealing one of the source images with the strange corruption.

I hope you can fix whatever bug caused this problem.

Many thanks,

Sandi

PS - out of curiosity, once I had the Focus Merge output, I wanted to open one of the original image files in Affinity Photo in a second tab to see that that file was fine but once I did, clicking back on the tab where the Focus Merge output was, I lost access to the panel with the source images. I could not find out how to bring that window of the source images back.

[John Rostron]

@Salus, I wonder if your images are misaligned. Or perhaps, as you change focus, the field of view alters. Did you bracket the photos  by changing the focus with the lens or by racking the whole camera to and fro?

John

[Salus]

Hi -

I took the images with my camera on a tripod and only varied the plane of focus by manually turning the focus dial slightly between each exposure. The camera itself and the item I photographed were both stationary and were not moved. 
 

Does that info help?

 

(I had done some other focus stacking today - 2 other stacks - on images taken the same day, on the same item and in the same way one right after the other with no problems.)

[Salus]

Also:

When looking at the source files after the focus merge was complete, the corrupt portion of the example image I included in my original post was found in the other source images as well. Or at least those that I looked at. (I didn’t review all 35 source images, just the first few...

[John Rostron]

Can I suggest that  you use a photo-viewing app to quickly view each image in turn? I suspect that the rogue areas are not always in the field of view, so cannot properly be combined in any focus merging algorithm. All I can suggest in this case is that you crop the final merge to exclude the rogue areas.

If you can increase the distance between the camera and the subject, using a longer focal length lens, it will give a more consistent field of view and reduce the chances of getting rogue areas. It would also help if you give some breathing space around your subject.

John

[lepr]

6 minutes ago, John Rostron said:

I suspect that the rogue areas are not always in the field of view,

Yes, the software is downscaling images to align their content with the one with the widest field of view. I use a macro lens with the same spec as the OP's lens - 90mm f/2.8 - and it suffers from focus breathing, too.

[Salus]

Thanks for the suggestions.

But I still do not understand why the source images do not look like the original images I used to make the stack - there is that weird effect at the edges that appears to be why there is a corresponding blur in the output merge image. This says to me that something is weird with the software - I mean why would my original images get changed during the process? Shouldn't the original images in the sources panel look exactly like the original images I used? Aligning the images should not change the content of the source image, should it? This is why I thought I should post on the forum - I thought maybe there was something in the program that was buggy that might need a fix. Am I thinking incorrectly about this? (I am a newbie so I certainly could be.)

 

Here again (attached image) is what one of the source images looks like (and it is not the only one!) after the focus merge has been implemented. That weird stuff on the side (less so on the bottom) is definitely not in the original image I input to the focus stack. Is this not a software problem in that the source images are getting altered?

 

Thanks for your time and expertise!

[John Rostron]

My guess is that the original source image was missing the area on the left, so the Focus merging algorithm made a guess as to what should be there and that is what you see in the 'source' as displayed. As I said, this foible is inherent in the focus merging process which is why I suggested you give your subject more room in the field of view.

John

[Salus]

Okay, I have been mulling over what @anon2 said - and should I think that the strange edges added to my source image are supposed to suggest that the image was scaled down to adjust for field of view and this is shown to me by this strange....(I do not know what to call it).....business at the edges?

 

Maybe I have finally gotten it. I am new to this and there isn't anything in the tutorials that says what the source image would look like if it must be aligned to the widest field of view.

 

In any case, I certainly will give myself some breathing room at the edges in future.

 

(Note: I've processed a few more of my stacks and so far have not seen this again. Yet. But I will be prepared to accept this explanation if I do.

 

Thanks.

 

[lepr]

1 hour ago, Salus said:

Aligning the images should not change the content of the source image, should it?

The thumbnails in your Sources panel reveal the field of view changing as the focus distance of the lens is changing, despite the lens having a fixed focal length of 90 mm. That phenomenon is known as focus breathing. In other words, the scale of each shot is changing and so the focus merging software has to perform compensatory scaling of each shot to align the shots.

[lepr]

5 minutes ago, Salus said:

this strange....(I do not know what to call it).....business at the edges

The boundary pixels of the downscaled content are being repeated to fill in the empty space.

[Salus]

Thanks to you both for all of the informative and helpful information. It’s nice to be made aware of how these processes are actually working. I feel a bit more enlightened now. Thanks again. 

[kirkt]

Take a look at using a focus rail to iterate your focus plane instead of refocusing the lens.

Kirk